1 Introduction

The rapid development of industrialization and urbanization has led to a continuous increase in carbon emissions. At present, carbon dioxide emissions from energy consumption are the main source of carbon emissions and the main cause of climate change. 30% to 50% of the total human greenhouse gas emissions come from carbon dioxide emitted from the construction sector, and carbon emission from construction sector is considered to be one of the main factors leading to global warming and environmental unsustainability [1]. According to statistics, the energy consumed by residential buildings worldwide accounts for 10% to 35% of the total energy consumption [2]. Cities are important units and research subjects for carbon emission reduction in countries and regions. The realization of national-level carbon emission reduction targets for buildings depends on the joint efforts of all cities.

2 Significance

As a prefecture-level city, Suzhou has always maintained rapid economic development and rapid urbanization. The construction sector consumes a lot of energy and also brings a lot of carbon emissions. Qualitative and quantitative research on the influencing factors of building carbon emissions in Suzhou will help reduce carbon emissions in the building sector and help achieve the carbon peak and neutrality goals. Taking Suzhou City as a typical case, it also has reference significance for the low-carbon development of other urban construction fields.

In the existing literature research, the influencing factors of building carbon emissions are mainly divided into four categories: (1) population. Glaeser&Kahn found that the carbon emission level of cities will increase with the increase of population through the research on the carbon emission of American cities [3]. EU research shows that the relationship between population and carbon emissions is related to the stage of regional economic development. Usually, in the early stage of economic development, carbon emissions increase with the increase of population, but the increase is decreasing [4]. (2) Income level. The research of domestic scholars shows that the carbon emissions of residents’ lives will be affected by their income. In low-income areas, carbon emissions are more sensitive to income, and the growth rate of carbon emissions of residents’ lives exceeds the growth rate of economic growth and residents’ income [5]. (3) Urbanization level. A large number of foreign studies show that the improvement of urbanization level will increase the total energy consumption, and thus increase the carbon emissions in the use of buildings [6, 7]. Sathaye believes that the improvement of urbanization level will increase the number of electrical appliances used by residents, thus increasing the total energy consumption [8]. The research of Chinese scholars shows that the lower the level of urbanization, the stronger the impact of urbanization on building carbon emissions [9]. (4) Per capita added value of the tertiary industry. The tertiary industry mainly refers to the service industry. Compared with the primary industry and the secondary industry, most of the tertiary industry dominated by the service industry occurs in construction sites. The increase of the added value of the tertiary industry means that the tertiary industry activities occurring in the construction sites increase, so the energy consumption and carbon dioxide emissions occurring in the construction sites will increase [10].

In analyzing the influencing factors of carbon emissions, the main methods used are Kaya identity decomposition, LMDI decomposition and IPAT model. The IPAT model is widely used in the research field of carbon emission influencing factors. However, the IPAT model has a major disadvantage, that is, the changes in the influencing factors of carbon dioxide emissions will be transferred to the carbon dioxide emissions in the same proportion [11]. The improved STIRPAT model can be used to study the non-proportional effects of various influencing factors on carbon emissions, which overcomes the above defects and is more widely used. Shahbaz et al. (2016) studied the influencing factors of urban carbon emissions based on the STIRPAT model, and found that the relationship between urbanization rate and urban carbon emissions is an inverted “U” shape [12]. Shao et al. (2010) improved the STIRPAT model and applied the improved model to study the influencing factors of carbon emissions in Shanghai [13]. Ding Weijia et al. (2012) applied the STIRPAT model to analyze the relationship and extent of the impact of three factors of population, wealth and technology on China’s manufacturing carbon emissions [14]. Niu Honglei and Liu Zhiyong applied the Dynamic Space Durbin panel data model and the STIRPAT model to analyze the impact of different factors on carbon emissions in the construction industry [15]. At present, most of the researches focus on the carbon emission evaluation of new buildings and the building use process, but the evaluation and analysis of the impact of carbon emissions in the main stages of the life cycle of urban buildings are still limited. Yang Yanfang (2016) analyzed the buildings in Beijing based on the STIRPAT model and studied the factors affecting carbon emissions [11]. Huang Zhenhua (2018) analyzed the buildings in Chongqing based on the STIRPAT model and studied the factors affecting carbon emissions [10]. To bridge this gap, this paper takes Suzhou City as an example and uses the improved STIRPAT model to analyze the influencing factors of carbon emissions in the main stages of the life cycle of urban buildings.

3 Research methods

3.1 STIRPAT model

The IPAT model was first proposed by Ehrlich & Holdren of Stanford University in the United States. It is mainly used to analyze the relationship between environmental pollution and economic development. The model is shown in Eq. (1):

$$\textit{I}=\textit{PAT}$$
(1)

The model has been widely used in the field of resource and environmental economics, but the model has a defect, that is, it is assumed that each factor of the model will be transmitted to the environment in equal proportions. The STIRPAT model for stochastic environmental impact assessment (Dietz and Rosa, 1994) [16] is based on the full application of IPAT theory, which is simple in concept, quantitative analysis, and comprehensive systematic characteristics, and further increases the randomness and non-uniformity between influencing factors and the environment. The model expression is shown in Eq. (2).

$$\textit{I}=\textit{aP}^\textit{b}\textit{A}^\textit{c}\textit{T}^\textit{d}\textit{e}$$
(2)

In the equation, I is the degree of environmental influence; P is the population influence factor; A is the economic influence factor; T is the technical influence factor; a is a constant term; b, c and d are the elastic coefficients of the impact of population, economy and technology on the environment respectively. e is the error term of the relation.

After linearizing the STIRPAT model, it can be further expressed as Eq. (3):

$$\ln\textit{I}=\textit{a}+\sum\nolimits_{\textit{i}=1}{\textit{b}}_\textit{i}\ln{\textit{P}}_\textit{i}+\sum\nolimits_{\textit{j}=1}{\textit{c}}_\text{j}\ln\;{\textit{A}}_\textit{j}+\sum\nolimits_{\textit{k}=1}{\textit{d}}_\text{k}\ln\;{\textit{T}}_\textit{k}+\textit{e}$$
(3)

In the equation, a is the constant coefficient term; e is the residual error; bi, cj, dk are the influence coefficients of each factor, the higher the value, the greater the influence of the corresponding factor on the environment.

3.2 STIRPAT model of building carbon emissions in Suzhou

This study takes Suzhou City as an example, and from the perspective of life cycle carbon emission analysis, the factor of urbanization level is introduced, and divides the economic influence factor into two factors: per capita disposable income and per capita added value of construction industry and tertiary industry in Suzhou. The technical factors are represented by the carbon emissions of the added value of the construction industry and the tertiary industry in Suzhou. The STIRPAT model of building carbon emissions in Suzhou established in this study is shown in Eq. (4).

$$\mathrm{lnC}=a+b\ln R+c\ln H+d\ln S+e\ln D+\mathrm g$$
(4)

In the formula, C is the carbon emission of buildings in Suzhou; R is the number of permanent residents in Suzhou (10,000 people); H is the urbanization rate of Suzhou (%); S is the per capita disposable income of Suzhou (10,000 yuan/person); D is the city’s per capita construction industry and tertiary industry added value (ten thousand yuan / person); a is the constant coefficient term; g is the residual error; b, c, d and e are the influence indices of each factor, respectively, see Eq. (8).

4 Data sources and processing

4.1 Calculation of building carbon emissions in Suzhou

Referring to “Building Carbon Emissions Calculation Standard” (GB/T51366–2019) [17], and Yang Yanfang's research [11], considering the availability of data, this paper defines building life cycle carbon emissions as the carbon dioxide of building spanning the preparation stage of building materials, carbon dioxide emissions in the building construction stage and carbon dioxide emissions in the building operational stage, and calculate the carbon dioxide emissions of buildings in Suzhou from 2010 to 2019. The data used in this paper come from the China Energy Statistical Yearbook [18], the China Construction Industry Statistical Yearbook [19], the Suzhou Statistical Yearbook [20] and the data released by relevant industry federations.

From the perspective of the whole life cycle, the CO2 emissions of buildings can be calculated by Eq. (5).

$$\textit{W}=\sum\nolimits_{\textit{i}=1}^3{\textit{E}}_\textit{i}$$
(5)

In the equation: W is the total carbon dioxide emissions in the whole life cycle of the building; E1 is the carbon dioxide emissions in the preparation stage of building materials; E2 is the carbon dioxide emissions in the building construction stage; E3 is the carbon dioxide emissions in the building use stage.

This study calculated the life cycle CO2 emissions of buildings in Suzhou for a total of 10 years from 2010 to 2019, as shown in Table 1. Fig. 1 shows the annual change of carbon dioxide emissions in the building life cycle in Suzhou from 2010 to 2019. As can be found in Fig. 1, the total carbon dioxide emissions from buildings in Suzhou increased from 54.71 million t in 2010 to 74.6 million t in 2019, and showed an overall upward trend, resulting a cumulative increase of 36.4% during this decade.

Table 1 Suzhou City’s carbon dioxide emissions during the building life cycle from 2010 to 2019
Full size table
Fig. 1
figure 1

Suzhou City’s carbon dioxide emissions during the building life cycle from 2010 to 2019

Full size image

4.1.1 CO2 emissions in the preparation stage of building materials

The carbon emissions in the preparation stage of building materials mainly refer to the carbon dioxide emissions during the production of building materials such as steel, aluminum, cement and other non-metallic building materials consumed in the building construction process. For a specific city, the carbon dioxide emissions generated by the materials of a single building can be obtained by multiplying the ratio of the city’s building area to the country’s total building area by the national carbon dioxide emissions from the materials of a single building. By synthesizing the industrial carbon dioxide emissions from the production of building materials, carbon dioxide emissions fron the urban building material preparation are obtained. The formula for calculating carbon dioxide emissions from a single building material in a single city is shown in Formula (6):

$$F_i=L_i\times M\times Z_i\times P_i$$
(6)

Among them, Fi is the carbon dioxide emission of building materials; Li is the comprehensive energy consumption of building material production; Zi is the building utilization ratio; Pi is the carbon emissions per unit of energy consumption in the production of building materials, Pi = (2.66x + 1.73y + 1.56z + 0.6101e)/(x + y + z + e), x, y, z, and e are the proportions of coal, oil, natural gas and electricity in the total final energy consumption of buildings; x, y, z, the coefficients in front of e are the carbon dioxide emission coefficients for coal, oil, natural gas and electricity, respectively. M is the ratio of the construction area of city to the total construction area of the country.

The energy consumption data of various types of building materials comes from the China Energy Statistical Yearbook [18], and the construction area data comes from the China Construction Industry Statistical Yearbook [19]. Since the ferrous metal smelting and rolling processing industry in the statistical data is mainly the production of steel, the energy consumption data of the ferrous metal smelting and rolling processing industry is used as the energy consumption data of steel production. The non-ferrous metal smelting and rolling processing industry in the statistical data includes not only the production of aluminum, but also lead, zinc, copper, etc. Therefore, the product of the proportion of energy consumption data of non-ferrous metal smelting and rolling processing industry and aluminum production in the total output of non-ferrous metal industry, is used as the energy consumption data of aluminum production. Since the non-metallic mineral product industry in the statistical data is mainly the production of building materials such as cement, the energy consumption data of the non-metallic mineral product industry is used as the energy consumption data for the production of non-metallic building materials such as cement. Zsteel is the ratio of steel used in buildings and is derived from relevant research report data as 55.4% [21]. ZAluminum is the use ratio of aluminum materials in buildings and set as 33% based on relevant research report data [22]. Table 2 shows the carbon dioxide emissions and related parameters in the preparation stage of building materials in Suzhou from 2010 to 2019 (Table 3).

Table 2 Carbon dioxide emissions and related parameters in the preparation stage of building materials in Suzhou from 2010 to 2019
Full size table
Table 3 Carbon dioxide emissions and related parameters in the construction stage of Suzhou City from 2010 to 2019
Full size table

4.1.2 CO2 emissions during the construction phase of buildings

The carbon dioxide emissions in the building construction stage are the total carbon dioxide emissions of the construction projects under construction in the whole society within one year. The calculation formula in this paper is shown as Eq. (7):

$$\textit{D}=\textit{H}\times\textit{M}\times\textit{P}$$
(7)

Among them, D is the carbon dioxide emission of the construction industry in Suzhou; H is the energy consumption of the national construction industry,from China Energy Statistics Yearbook [18]; M is the ratio of the construction area of Suzhou to the total construction area of the country, from China Construction Industry Statistics Yearbook [19] and Suzhou Statistics Yearbook [20]; P is the carbon dioxide emission per unit of energy consumption, and P = (2.66x + 1.73y + 1.56z + 0.6101e)/ (x + y + z + e), x, y, z, and e are the proportions of coal, oil, natural gas and electricity in the total final energy consumption of buildings; the coefficients in front of x, y, z, and e are CO2 emission factors for coal, oil, gas and electricity, respectively.

4.1.3 CO2 emissions during the building operational phase

The carbon dioxide emissions in the building operation stage are mainly carbon dioxide emissions caused by the existing building air conditioning, lighting, domestic hot water supply, cooking utensils and other building energy equipment such as electricity, natural gas, liquefied petroleum gas and other energy sources, as shown in Table 4.

Table 4 Carbon dioxide emissions of buildings in Suzhou during 2010–2019
Full size table

4.2 Data collection of building carbon emissions influencing factors

The influencing factors of building carbon emissions in Suzhou include the number of permanent residents in Suzhou (R), urbanization rate (H), per capita disposable income (S), per capita construction and tertiary industry added value (D), see Table 5.

Table 5 Data of factors affecting building carbon emissions in Suzhou from 2010 to 2019
Full size table

5 Analysis of factors affecting building carbon emissions

5.1 Variable multicollinearity diagnosis of factors influencing building carbon emissions

Multicollinearity among variables means that there is a linear correlation between variables. When the variables have severe multicollinearity, the stability of the regression parameter estimates obtained by the ordinary least squares regression method is reduced, and the coefficient t test cannot pass [23, 24]. The variance inflation factor (VIF) can be used to diagnose the multicollinearity of the variables [25]. If VIFJ ≥ 10, it means that the variables in the model have strong collinearity. In this study, SPSS 20.0 is used to conduct least squares regression analysis [26, 27] and multicollinearity VIF test on the STIRPAT model (4) of the influencing factors of building carbon emissions in Suzhou constructed in this study. The results are shown in Table 6. The VIF values of each factor in Table 6 are all greater than 10, and the maximum value is 3486.044, so it can be judged that there is serious multicollinearity among the variables.

Table 6 The results of the least squares regression method for diagnosing multicollinearity
Full size table

5.2 Ridge regression

In this paper, a ridge regression method is used to eliminate multicollinearity in the model [28, 29]. Use SPSS 20.0 to perform ridge regression fitting to obtain a ridge trace diagram, as shown in Fig. 2. It can be seen from Fig. 2 that when k = 0.13, the ridge traces of the five independent variables in Fig. 2 tend to be stable. Therefore, in this study, k = 0.13 is selected as the ridge value, and the linear regression results based on ridge regression are obtained. The specific results are shown in Table 7.

Fig. 2
figure 2

The regression ridge trace of the analysis model of the influencing factors of building carbon emissions in Suzhou

Full size image
Table 7 Ridge regression linear regression results
Full size table

It can be seen from Table 7 that the coefficient of determination R-Square of the ridge regression model is 0.9945, which has a high goodness of fit. In the hypothesis test of the ridge regression model, F-value = 101.7678, the test significance p < 0.0001, and the analysis model of the influencing factors of building carbon emissions in Suzhou City is shown in Eq. (8).

$$\text{lnC}=5.2959+0.2701\text{ln}\textit{R}+0.3051\text{ln}\textit{H}+0.0665\text{ln}\textit{S}+0.0778\text{ln}\textit{D}$$
(8)

From the coefficients of each independent variable in Formula (8), it can be seen that,all factors have a positive impact on building carbon emissions. The degree of influence of each factor on building carbon dioxide emissions in Suzhou is in descending order: urbanization rate, number of permanent residents, per capita construction and tertiary industry added value, and per capita disposable income. The elastic coefficients are 0.3051, 0.2701, 0.0778 and 0.0665.

5.3 Analysis of model results

The diagnostic results of multicollinearity indicate that the respective variables interacted with each other. The analysis is as follows: Suzhou’s industrialization has developed rapidly, urban construction has been accelerated, and a large number of migrants have been attracted, and the number of permanent residents has grown rapidly. The improvement of urbanization also brings about changes in the urban and rural structure of the population, triggering changes in production and consumption behavior, and superimposing the increase in population, which promotes the rapid increase in the demand for various service products of residents, and drives the rapid development of the construction industry and the tertiary industry. At the same time, residents’ income increased and residents’ consumption level improved, which further promote the development of the service industry [5].

The results of the ridge regression model analysis show that the urbanization rate has the most significant impact on the carbon emissions of buildings in Suzhou, and its elasticity coefficient reaches 0.3051, indicating that the urbanization rate is one of the main long-term driving factors for carbon emissions in the building sector in Suzhou. The improvement has strongly driven the increase of building carbon emissions [9]. The main reason is that the urban construction increases the number of buildings and promotes the growth of total carbon emissions from buildings. The second is the resident population of Suzhou, whose elasticity coefficient reaches 0.2701, indicating that there is a long-term driving relationship between the resident population of Suzhou and the carbon emissions of the building sector in Suzhou. The main reason is that the growth of the resident population leads to the demand for residential buildings and public service buildings. The rapid development of the housing construction industry and the increase in various activities and energy consumption in construction sites have led to an increase in the total amount of building carbon emissions. Again, it is the added value of the construction industry and the tertiary industry per capita, and its elasticity coefficient is 0.0778. The construction industry is the housing construction industry of the whole society, which represents the development level of urban buildings, and the growth of its added value brings an increase in carbon emissions during the construction phase. Compared with the primary industry and the secondary industry, most of the tertiary industry dominated by the service industry occurs in construction sites, such as catering, accommodation, shopping malls, etc. when the added value of the tertiary industry increases, it will inevitably lead to increase activity and energy consumption in public areas, which in turn leads to increase carbon emissions during the building’s operational phase. The last is per capita disposable income, and its elasticity coefficient is 0.0655. The growth of per capita disposable income means the improvement of people’s living standards, the increase of residents’ living energy consumption, and the increase of carbon emissions from residential buildings. Building carbon emission intensity is the proportion of building carbon emissions to the total added value of the construction industry and the tertiary industry.

6 Conclusion

In summary, industrialization and urbanization have strongly promoted the growth of building carbon emissions in Suzhou. In the future, with the continuous development of industrialization and urbanization and the continuous growth of the population, Suzhou City can rationally plan the urban development boundary to promote the green and low-carbon transformation and development of urban and rural construction, cultivate residents’ low-carbon awareness, and increase residents’ green and low-carbon behaviors to reduce building carbon emissions. The specific suggestions are as follows: the first is to promote green construction, promote green low-carbon building materials, develop prefabricated buildings, promote the integrated construction of photovoltaic buildings, promote the optimization of building energy consumption structure, promote the energy-saving transformation of existing buildings and municipal infrastructure in cities and towns, and promote the organic renewal of cities. The second is to establish an incentive mechanism for the public to participate in green and low-carbon, implement green life creation actions, improve the green and low-carbon awareness of the whole people, advocate a green and low-carbon lifestyle, and promote green consumption.